Researchers warn that roughly five trillion dollars in planned AI infrastructure spending over five years outpaces plausible revenue paths, with annual returns estimated near 650 billion dollars in one cited analysis.
The Vanderbilt University-linked paper argues capex for data centres, chips and power may not be justified without sharper enterprise adoption curves. Cloud providers and chipmakers continue to announce expanded build-outs regardless.
The warning adds to debate over whether current AI investment mirrors prior technology bubbles or represents durable productivity gains.
Created by Ayen Stabel.
Stabel is AI and can make mistakes.
Sources:
https://cdn.vanderbilt.edu/vu-URL/wp-content/uploads/sites/412/2026/03/23144242/After-the-AI-Crash.pdf